Latent Sculpting for Zero-Shot Generalization: A Manifold Learning Approach to Out-of-Distribution Anomaly Detection

Avinash Kumar; Rajeeb Thapa Chhetri; Saurab Thapa; Zhixiong Chen

REVIEW 4 major objections 5 minor 18 references

By sculpting benign traffic into a compact latent cluster before density estimation, this framework claims zero-shot detection of unseen network attacks at an F1 of 0.87.

Reviewed by Pith at T0; open to challenge. T0 means a machine referee read the full paper against a public rubric. the ladder, T0–T4 →

Challenge this review Re-run · record.json Download PDF Read on arXiv ↗

T0 review · deepseek-v4-flash

2026-08-03 15:14 UTC pith:5F6HLHNR

load-bearing objection A promising two-stage OOD pipeline with a confounded central claim; worth a careful revision, not a desk reject. the 4 major comments →

arxiv 2512.22179 v3 pith:5F6HLHNR submitted 2025-12-19 cs.LG cs.CR

Latent Sculpting for Zero-Shot Generalization: A Manifold Learning Approach to Out-of-Distribution Anomaly Detection

Rajeeb Thapa Chhetri , Saurab Thapa , Avinash Kumar , Zhixiong Chen This is my paper

classification cs.LG cs.CR

keywords zero-shot generalizationout-of-distribution detectiontabular anomaly detectionmanifold learningnormalizing flowsnetwork intrusion detectionCIC-IDS-2017Dual-Centroid Compactness Loss

verification ladder T0 review T1 audit T2 compute T3 formal T4 reserved

The pith

A machine-rendered reading of the paper's core claim, the machinery that carries it, and where it could break.

The paper tries to establish that explicit “latent sculpting”—compressing benign network traffic into a tight, separated cluster during representation learning—is what makes density-based zero-shot anomaly detection work. Its core claim is that decoupling structure learning (Stage 1) from density estimation (Stage 2) resolves the “generalization collapse” of supervised detectors, which in a prior baseline study dropped to F1≈0.30 on unseen attacks. On the CIC-IDS-2017 benchmark, the two-stage system reports a zero-shot out-of-distribution F1 of 0.87 and AUROC of 0.913 at a high-sensitivity threshold, with the isolated Stage-1 encoder reaching only 8% recall—a gap the authors present as proof that the sculpting, not the geometry alone, enables detection. A sympathetic reader would care because the method promises a practical blueprint for detecting zero-day cyber threats without retraining, using a fraction of the benign data of prior supervised models.

Core claim

The central discovery is that the latent space's topology—not the architecture alone—determines whether a subsequent normalizing flow can separate novel anomalies from normal data. Stage 1 trains a hybrid 1D-CNN and Transformer encoder with a Dual-Centroid Compactness Loss (DCCL), which pulls benign samples toward a benign centroid, pulls known-anomaly samples toward an anomaly centroid, and pushes the two centroids apart beyond a margin m=5.0. Stage 2 freezes this encoder and fits a Masked Autoregressive Flow solely on the benign embeddings, so that benign data maps to a standard Gaussian while anything outside the compact manifold maps to low-probability tails. On held-out attack classes (

What carries the argument

The load-bearing object is the Dual-Centroid Compactness Loss (DCCL), a loss that enforces absolute topological constraints: it minimizes intra-cluster variance around a benign centroid and an anomaly centroid while maximizing the Euclidean distance between those centroids, creating a high-density “cliff” around normality. The second component is the Masked Autoregressive Flow, a normalizing flow that learns an invertible map from the compact benign latent cluster to a standard Gaussian; the log-likelihood of a test point under this map becomes the anomaly score. Together they implement the paper's “structure-then-estimate” principle: DCCL manufactures the compact geometry, and the flow expl

Load-bearing premise

The method assumes that compacting the benign cluster with DCCL pushes every meaningful zero-day attack outside the high-probability region of the Stage-2 flow; if an unseen attack projects inside that cluster, it is missed—and the paper's own Bot result (4.07% recall) shows this premise does not always hold.

What would settle it

A concrete observation that would settle the claim: the paper's own Bot class, which mimics benign HTTP keep-alive behavior, is detected at only 4.07% recall—an explicit violation of the assumption that all unseen attacks land outside the compact benign cluster. A direct experiment would ablate DCCL (train Stage 1 with a plain contrastive or classification loss, keep Stage 2 identical) and check whether the 8%-vs-85% recall gap collapses; if it does, the sculpting loss is the causal ingredient, and if it does not, the two-stage recipe alone is responsible.

Watch this falsifier — get emailed when new claim-graph text bears on it.

If this is right

If correct, a compact latent manifold plus normalizing-flow density is sufficient to flag zero-day attacks that supervised classifiers miss, including stealthy infiltration attacks that drove supervised baselines to 0% accuracy.
The framework could bridge supervised and unsupervised anomaly detection, matching supervised precision on known attacks (F1≈0.96) while exceeding unsupervised baselines on unseen ones (F1 0.87 vs 0.76).
Data efficiency: only about 10% of the benign traffic used by prior supervised baselines sufficed for training, which would lower the cost of adapting to new network environments.
Thresholding on NLL percentiles (P99/P97/P95) yields a tunable sensitivity spectrum, letting an operator trade false alarms against catching low-magnitude attacks.
The 8%-vs-85% recall gap between single-stage and two-stage inference indicates that distance to centroids alone is inadequate, and probabilistic density is necessary for OOD rejection.

Where Pith is reading between the lines

These are editorial extensions of the paper, not claims the author makes directly.

The framework's failure on Bot traffic (4.07% recall) points to a boundary condition: attacks that mimic benign semantics in the latent space will evade any purely geometric/density approach; a temporal or behavioral module would be needed, a direction the paper itself flags in its future-work section.
Extending beyond network data, the same “structure-then-estimate” recipe could be tested on visual or medical anomalies by swapping the 1D-CNN front-end for a 2D backbone, as the authors propose; the near-total Bot miss suggests the transfer would inherit the same blind spot for semantically normal-looking outliers.
Reproducing the 8%-vs-85% gap with an ablated encoder (e.g., plain contrastive loss without DCCL) would be a direct test of whether the compactness loss, rather than the two-stage recipe, is the causal ingredient.
A reader should confirm the precise operating point behind the headline numbers, since the abstract reports AUROC 0.913 while the results table lists 0.9023 at the P99 threshold.

Editorial analysis

A structured set of objections, weighed in public.

Desk editor's note, referee report, simulated authors' rebuttal, and a circularity audit.

Desk Editor's Note

A promising two-stage OOD pipeline with a confounded central claim; worth a careful revision, not a desk reject.

read the letter

Short version: this is a clean, well-documented two-stage pipeline that reports a strong zero-shot OOD F1 of 0.87 on CIC-IDS-2017, but the paper’s central claim — that ‘latent sculpting’ is what makes it work — is not actually tested. The key experiment compares an 8% recall distance-based scorer to an 85% recall flow-density scorer, which changes the scoring function at the same time as adding the flow. The authors’ own LOF baseline gets 83.7% recall without any sculpting, so the jump is at least as easily explained by density scoring itself. That is a load-bearing confound.

What is genuinely here: a concrete architecture (1D-CNN + Transformer + DCCL + MAF), a careful seen/unseen split, and unusually honest reporting of limitations and AI use. The Infiltration result (88.89% on a class that supervised baselines miss even when trained on it) is worth a look, and the balanced internal validation is a fair standard.

The soft spots are real but addressable. First, there is no ablation that isolates DCCL: I want to see the same two-stage pipeline with an encoder trained without DCCL (or with plain cross-entropy or center loss only). Without that, the ‘sculpting is a prerequisite’ claim is a counterfactual, not a result. Second, the threshold is not zero-shot: the authors sweep three percentiles and report the one that performs best on the OOD test set. The protocol should fix the threshold from the training set alone. Third, the paper’s numbers do not line up: the abstract says an 85th-percentile threshold and 78.7% Infiltration recall; the body’s tables say P95 and 88.89%. Fourth, the Bot class sits at 4% recall; the authors acknowledge it, but it does contradict the overall ‘stable zero-day detection’ framing. None of these is fatal, but the central argument needs to be reframed as a hypothesis plus promising evidence, not a proof.

This paper deserves a serious referee. It is not a desk reject: the method is concrete, the evaluation is more careful than average, and the flaws are fixable. I would ask the authors to add a DCCL ablation, pre-specify the threshold, and reconcile the abstract with the tables. A reader interested in OOD detection for network traffic will get a useful case study in how easy it is to overclaim from a confounded comparison.

Referee Report

4 major / 5 minor

Summary. The manuscript proposes a two-stage framework called Latent Sculpting for zero-shot OOD anomaly detection on tabular network traffic. Stage 1 trains a hybrid 1D-CNN/Transformer encoder using a Dual-Centroid Compactness Loss (DCCL) to pull benign samples into a compact cluster and separate it from known attack clusters. Stage 2 freezes the encoder and fits a Masked Autoregressive Flow (MAF) to the benign embeddings, then flags samples by NLL against percentile thresholds. On CIC-IDS-2017, the authors report F1=0.87/recall=0.85 at the P95 threshold for four unseen attack classes, outperforming OCSVM (F1=0.7575) and LOF (F1=0.6814) from Xu et al., and they interpret the Stage-1-vs-full-pipeline gap as evidence that explicit latent sculpting is a prerequisite for density-based OOD detection.

Significance. If the causal claim were established, the 'structure-then-estimate' recipe would be a practical contribution to tabular anomaly detection. The paper is transparent about computational reproducibility (fixed seed, full hyperparameter table, code URL) and candidly reports failure cases such as Bot, Web attacks, and SSH-Patator. The data-efficiency claim (training on ~184k benign samples instead of ~1.8M) is also interesting. However, the current experiments do not isolate the effect of DCCL, and the headline numbers in the abstract and body are inconsistent, so the significance of the central claim is not yet verifiable.

major comments (4)

[§6.2, §8.1 (Tables 6, 8)] The paper's central evidence for latent sculpting is the contrast between the isolated Stage 1 encoder (8% recall, Euclidean distance scoring) and the full two-stage pipeline (85% recall, MAF NLL scoring). This comparison changes two variables at once — the presence of Stage 2 and the anomaly scoring function. It therefore cannot isolate the effect of DCCL. The paper's own baseline LOF (Table 8) achieves 83.7% recall on the same OOD set with no sculpting at all, which shows that density-based scoring alone nearly matches the reported result. Without an ablation that removes or weakens DCCL while keeping Stage 2 fixed, the statement that 'DCCL is a prerequisite' is an untested counterfactual, not evidence. Please add such an ablation or reframe the contribution as a pipeline-level result.
[Abstract vs §4.2.1, Tables 6–7] The headline numbers are inconsistent. The abstract reports a 'balanced threshold (85th-percentile)' with zero-shot OOD F1=0.867 and AUROC=0.913, and Infiltration recall of 78.7% peaking at 97.2%. However, §4.2.1 defines the balanced threshold as P97, not the 85th percentile; Table 6 gives P97 F1=0.74/AUROC=0.9023, and P95 (the only row with F1=0.87) is called 'Sensitive'. Table 7 reports Infiltration recalls of 69.44%/86.11%/88.89% across P99/P97/P95, with no 78.7% or 97.2% anywhere. The abstract and conclusion must be aligned with the body before the results can be evaluated.
[§3.2, §6.3, Table 7] The method's key assumption — that 'any z vector falling even slightly outside this dense manifold will result in a u vector far from the Gaussian's center' — is load-bearing but empirically false for at least one OOD class. Bot traffic has only 4.07% recall at P95, and among seen attacks SSH-Patator drops to 47.83% even at P99. Thus unseen attacks can project inside the compact benign cluster, and the claim that sculpting guarantees a probability cliff is too strong. This does not invalidate the pipeline, but it should be presented as a stated boundary condition, with class-wise scores (not only pooled F1) emphasized in the conclusion.
[§7.2/Table 8] The comparison to the unsupervised baselines is informative, but the 'state-of-the-art' framing depends on one selected operating point (P95) chosen after evaluating three thresholds. Since P97 gives F1=0.74 and P99 gives F1=0.67, the robustness of the claimed superiority over OCSVM (F1=0.7575) is not established at the other operating points. Please report the full threshold sweep in the comparison table and discuss how the operating point should be chosen in practice.

minor comments (5)

[Table 6] AUROC/AUPRC appear only on the P99 row; clarify whether these are threshold-specific or model-wide metrics.
[Tables 7–9] Some rows refer to '≈8%' aggregate recall for Stage 1 with an asterisk; this number is already in Table 6 and should not be repeated in per-attack comparisons.
[Throughout] The loss name is introduced as Dual-Centroid Compactness Loss (DCCL) but later written 'DualCompactnessLoss'; unify terminology.
[§4.1.1] Figure 1 shows MaxPool output as (B,16,33), but the text says the sequence length after convolutions is 66; with k=2 and no padding this should be 66 only if the input length permits it. Confirm the shape calculation.
[References] Reference [15] is cited both for the dataset and for the 'unseen threat landscape' in §6.1; consider a dedicated protocol citation if available.

Circularity Check

0 steps flagged

No significant circularity: zero-shot OOD results are not forced by construction.

full rationale

The derivation chain is self-contained in the relevant sense. Stage 1 (DCCL) is trained on benign plus seen-attack samples with the four OOD classes (Slowloris, Slowhttptest, Bot, Infiltration) excluded; Stage 2 MAF is trained only on benign latent embeddings; anomaly scores are NLL; thresholds are percentiles of benign training NLL (P99/P97/P95), not fit to OOD labels. Thus the reported OOD F1/AUROC are measured outcomes, not definitional consequences. The Stage-1-versus-Stage-2 comparison (8% vs 85% recall) is confounded because it changes the scoring function from Euclidean distance to MAF density, and the 'sculpting is a prerequisite' claim would need an ablation without DCCL; however, that is a correctness/experimental-design threat, not a circularity under the stated criteria. No load-bearing self-citation exists: baselines come from external Xu et al. [1], and the architectural components cite standard external work (Center Loss [7], MAF [6]). The threshold sweep across three percentiles is a mild selection concern, but the thresholds themselves are computed from benign training-data NLL, so the zero-shot result is not equivalent to an input by construction.

Axiom & Free-Parameter Ledger

4 free parameters · 4 axioms · 0 invented entities

The central machinery is a weighted sum of known losses plus a normalizing flow; the only invented objects are the loss weighting, margin, and thresholds, none of which have independent evidence. No new physical/structural entities are postulated.

free parameters (4)

DCCL loss weights α, β, γ = 0.1, 0.1, 1.0
Set by hand ("empirically tuned") to prioritize separation; affects cluster geometry that Stage 2 relies on.
Separation margin m = 5.0
Hand-chosen Euclidean margin in L_s; directly creates the 'probability cliff' the authors attribute to the method.
Percentile thresholds τ99, τ97, τ95 = -164.22, -173.08, -177.19 (NLL)
Derived from training benign NLL; P95 is used for the headline and appears selected after inspecting test performance.
Latent dimension = 32
Architecture choice; compactness and density estimation are sensitive to it, but no sweep is reported.

axioms (4)

standard math Change-of-variables formula and invertibility of MAF layers
Used in Eq. 6-7 to convert latent density to Gaussian base density; standard result from Papamakarios et al. [6].
domain assumption CIC-IDS-2017 is a valid proxy for non-stationary network data and its train/test split has no leakage
The entire evaluation rests on this benchmark; no duplicate-flow removal is reported despite known duplicates in CIC-IDS-2017.
domain assumption Unseen attacks will fall in low-density regions of the compacted benign manifold
Central sculpting hypothesis; contradicted by Bot class, so it is an assumption, not an established fact.
domain assumption Stage 1's known-anomaly cluster is sufficiently representative to shape the latent space for arbitrary future attacks
Uses only seen attacks to sculpt; whether that transfers to unseen attack semantics is exactly what is being assumed.

pith-pipeline@v1.3.0-alltime-deepseek · 17325 in / 12613 out tokens · 121457 ms · 2026-08-03T15:14:10.777292+00:00 · methodology

0 comments

read the original abstract

Detecting previously unseen attacks remains a major challenge for machine learning-based intrusion detection systems. Deep models trained on network traffic often achieve high accuracy on known attacks but fail under distributional shift because their decision boundaries are tightly coupled to the training data distribution. We introduce Latent Sculpting, a two-stage anomaly detection framework that improves robustness by explicitly structuring the latent representation before density estimation. The first stage trains a Transformer-based tabular encoder using a novel Binary Latent Sculpting loss, which encourages benign traffic to form a compact latent cluster while enforcing separation from anomalous patterns. The second stage fits a Masked Autoregressive Flow to the resulting latent space to produce calibrated probabilistic anomaly scores. Under a strict zero-shot evaluation protocol on the CIC-IDS-2017 benchmark, Stage 1 attains an F1-score of 0.98 on known attacks, while Stage 2 -- evaluated at the balanced threshold (85th-percentile) -- achieves a zero-shot OOD F1-score of 0.867 and AUROC of 0.913. The model successfully detects difficult distribution shifts including stealthy infiltration attacks (78.7% recall, peaking at 97.2%) and low-volume DoS variants (>94% recall), scenarios where conventional approaches often fail. Our results suggest that explicitly separating latent geometry learning from density modeling provides a stable approach for detecting zero-day cyber threats.

Figures

Figures reproduced from arXiv: 2512.22179 by Avinash Kumar, Rajeeb Thapa Chhetri, Saurab Thapa, Zhixiong Chen.

**Figure 1.** Figure 1: 1D-CNN Feature Extractor Architecture. The model accepts an input of shape (B, 71), reshapes it, and processes it through five 1D convolutional layers with kernel size k = 2. ReLU activations provide non-linearity, while Dropout (p = 0.1) and MaxPool operations ensure robustness and dimensionality reduction before passing features to the Transformer stage. • Mechanism: – Structure: The network consists of … view at source ↗

**Figure 2.** Figure 2: Transformer Encoder and Latent Projection Head. The feature sequence (B, 16, 33) extracted by the CNN is permuted and projected to the Transformer embedding dimension dmodel = 64. After adding positional embeddings, the sequence is processed by a Transformer Encoder layer utilizing Multi-Head Attention and a Feed-Forward Network. Finally, Global Average Pooling aggregates the temporal dimension, and an MLP… view at source ↗

**Figure 3.** Figure 3: Masked Autoregressive Flow (MAF) Architecture. The density estimation stage consists of a stack of 16 MAF layers. Each layer receives the latent vector zenc from the Transformer Encoder (Stage 1) and transforms it through a Masked Autoencoder for Distribution Estimation (MADE). The MADE network outputs scale (s) and shift (t) parameters, which are used to apply an affine transformation u = z ⊙ exp(tanh(s))… view at source ↗

discussion (0)

Reference graph

Works this paper leans on

18 extracted references · 1 canonical work pages

[1]

Robust Anomaly Detection in Network Traffic: Evaluating Machine Learning Models on CICIDS2017.arXiv preprint arXiv:2506.19877v2, 2025

Xu, Zhaoyang and Liu, Yunbo. Robust Anomaly Detection in Network Traffic: Evaluating Machine Learning Models on CICIDS2017.arXiv preprint arXiv:2506.19877v2, 2025. Available: https://arxiv.org/abs/ 2506.19877

Pith/arXiv arXiv 2025
[2]

Outside the Closed World: On Using Machine Learning for Network Intrusion Detection.2010 IEEE Symposium on Security and Privacy, pages 305–316, 2010

Sommer, Robin and Paxson, Vern. Outside the Closed World: On Using Machine Learning for Network Intrusion Detection.2010 IEEE Symposium on Security and Privacy, pages 305–316, 2010. https://doi.org/10.1109/ SP.2010.25

2010
[3]

Deep Learning for Anomaly Detection: A Survey.arXiv preprint arXiv:1901.03407, 2019.https://arxiv.org/abs/1901.03407

Chalapathy, Raghavendra and Chawla, Sanjay. Deep Learning for Anomaly Detection: A Survey.arXiv preprint arXiv:1901.03407, 2019.https://arxiv.org/abs/1901.03407

Pith/arXiv arXiv 1901
[4]

Anomaly-Based Network Intrusion Detection: Techniques, Systems and Challenges.Computers & Security, 28(1-2):18–28, 2009

Garcia-Teodoro, Pedro, Diaz-Verdejo, Jesus, Maciá-Fernández, Gabriel, and Vázquez, Enrique. Anomaly-Based Network Intrusion Detection: Techniques, Systems and Challenges.Computers & Security, 28(1-2):18–28, 2009. https://doi.org/10.1016/j.cose.2008.08.003

doi:10.1016/j.cose.2008.08.003 2009
[5]

Outlier Detection with Autoencoder Ensembles

Chen, Jing, Sathe, Saket, Aggarwal, Charu, and Turaga, Deepak. Outlier Detection with Autoencoder Ensembles. Proceedings of the 2017 SIAM International Conference on Data Mining, pages 90–98, 2017. https://doi. org/10.1137/1.9781611974973.11

work page doi:10.1137/1.9781611974973.11 2017
[6]

Masked Autoregressive Flow for Density Estimation

Papamakarios, George, Pavlakou, Theo, and Murray, Iain. Masked Autoregressive Flow for Density Estimation. Advances in Neural Information Processing Systems (NeurIPS), 30:2335–2344, 2017. https://proceedings. neurips.cc/paper/2017/hash/6c150ed4-Paper.pdf

2017
[7]

A Discriminative Feature Learning Approach for Deep Face Recognition.European Conference on Computer Vision (ECCV), pages 499–515

Wen, Yandong, Zhang, Kaipeng, Li, Zhifeng, and Qiao, Yu. A Discriminative Feature Learning Approach for Deep Face Recognition.European Conference on Computer Vision (ECCV), pages 499–515. Springer, 2016. https://doi.org/10.1007/978-3-319-46478-7_31

doi:10.1007/978-3-319-46478-7_31 2016
[8]

Dimensionality Reduction by Learning an Invariant Mapping

Hadsell, Raia, Chopra, Sumit, and LeCun, Yann. Dimensionality Reduction by Learning an Invariant Mapping. 2006 IEEE Computer Society Conference on Computer Vision and Pattern Recognition (CVPR’06), 2:1735–1742, 2006.https://doi.org/10.1109/CVPR.2006.100

doi:10.1109/cvpr.2006.100 2006
[9]

Schroff, Florian, Kalenichenko, Dmitry, and Philbin, James. FaceNet: A Unified Embedding for Face Recognition and Clustering.Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition (CVPR), pages 815–823, 2015.https://doi.org/10.1109/CVPR.2015.7298682

arXiv 2015
[10]

Wang, Feng, Deng, Jian, Liu, Weiyang, and Wang, C. AMC-Loss: Angular Margin Contrastive Loss for Insightful Representation Learning.Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition Workshops (CVPRW), pages 3659–3666, 2020

2020
[11]

LossTransform: Reformulating Contrastive Objectives for Robust Representation Learning.International Conference on Learning Representations (ICLR), 2025

Liu, Z., Chen, X., and Zhang, H. LossTransform: Reformulating Contrastive Objectives for Robust Representation Learning.International Conference on Learning Representations (ICLR), 2025

2025
[12]

Attention Is All You Need.Advances in Neural Information Processing Systems (NeurIPS), 30:5998–6008, 2017.https://arxiv.org/abs/1706.03762

Vaswani, Ashish, Shazeer, Noam, Parmar, Niki, Uszkoreit, Jakob, Jones, Llion, Gomez, Aidan N, Kaiser, Łukasz, and Polosukhin, Illia. Attention Is All You Need.Advances in Neural Information Processing Systems (NeurIPS), 30:5998–6008, 2017.https://arxiv.org/abs/1706.03762

Pith/arXiv arXiv 2017
[13]

Representation Learning: A Review and New Perspectives

Bengio, Yoshua, Courville, Aaron, and Vincent, Pascal. Representation Learning: A Review and New Perspectives. IEEE Transactions on Pattern Analysis and Machine Intelligence, 35(8):1798–1828, 2013. https://doi.org/ 10.1109/TPAMI.2013.50 20 APREPRINT- DECEMBER30, 2025

doi:10.1109/tpami.2013.50 2013
[14]

A Baseline for Detecting Misclassified and Out-of-Distribution Examples in Neural Networks.Proceedings of the International Conference on Learning Representations (ICLR), 2017

Hendrycks, Dan and Gimpel, Kevin. A Baseline for Detecting Misclassified and Out-of-Distribution Examples in Neural Networks.Proceedings of the International Conference on Learning Representations (ICLR), 2017. https://arxiv.org/abs/1610.02136

Pith/arXiv arXiv 2017
[15]

Sharafaldin, Iman, Lashkari, Arash Habibi, and Ghorbani, Ali A. Toward Generating a New Intrusion Detection Dataset and Intrusion Traffic Characterization.Proceedings of the 4th International Conference on Information Sys- tems Security and Privacy (ICISSP), pages 108–116, 2018. https://doi.org/10.5220/0006639801080116

doi:10.5220/0006639801080116 2018
[16]

A Simple Framework for Contrastive Learning of Visual Representations.International Conference on Machine Learning (ICML), pages 1597–1607

Chen, Ting, Kornblith, Simon, Norouzi, Mohammad, and Hinton, Geoffrey. A Simple Framework for Contrastive Learning of Visual Representations.International Conference on Machine Learning (ICML), pages 1597–1607. PMLR, 2020.https://arxiv.org/abs/2002.05709

Pith/arXiv arXiv 2020
[17]

Decoupled Weight Decay Regularization.International Conference on Learning Representations (ICLR), 2019.https://arxiv.org/abs/1711.05101

Loshchilov, Ilya, and Hutter, Frank. Decoupled Weight Decay Regularization.International Conference on Learning Representations (ICLR), 2019.https://arxiv.org/abs/1711.05101

Pith/arXiv arXiv 2019
[18]

Data Preprocessing for Supervised Learning.International Journal of Computer Science, 1(2):111–117, 2006

Kotsiantis, Sotiris B., Kanellopoulos, Dimitris, and Pintelas, Panayiotis E. Data Preprocessing for Supervised Learning.International Journal of Computer Science, 1(2):111–117, 2006. 21

2006

[1] [1]

Robust Anomaly Detection in Network Traffic: Evaluating Machine Learning Models on CICIDS2017.arXiv preprint arXiv:2506.19877v2, 2025

Xu, Zhaoyang and Liu, Yunbo. Robust Anomaly Detection in Network Traffic: Evaluating Machine Learning Models on CICIDS2017.arXiv preprint arXiv:2506.19877v2, 2025. Available: https://arxiv.org/abs/ 2506.19877

Pith/arXiv arXiv 2025

[2] [2]

Outside the Closed World: On Using Machine Learning for Network Intrusion Detection.2010 IEEE Symposium on Security and Privacy, pages 305–316, 2010

Sommer, Robin and Paxson, Vern. Outside the Closed World: On Using Machine Learning for Network Intrusion Detection.2010 IEEE Symposium on Security and Privacy, pages 305–316, 2010. https://doi.org/10.1109/ SP.2010.25

2010

[3] [3]

Deep Learning for Anomaly Detection: A Survey.arXiv preprint arXiv:1901.03407, 2019.https://arxiv.org/abs/1901.03407

Chalapathy, Raghavendra and Chawla, Sanjay. Deep Learning for Anomaly Detection: A Survey.arXiv preprint arXiv:1901.03407, 2019.https://arxiv.org/abs/1901.03407

Pith/arXiv arXiv 1901

[4] [4]

Anomaly-Based Network Intrusion Detection: Techniques, Systems and Challenges.Computers & Security, 28(1-2):18–28, 2009

Garcia-Teodoro, Pedro, Diaz-Verdejo, Jesus, Maciá-Fernández, Gabriel, and Vázquez, Enrique. Anomaly-Based Network Intrusion Detection: Techniques, Systems and Challenges.Computers & Security, 28(1-2):18–28, 2009. https://doi.org/10.1016/j.cose.2008.08.003

doi:10.1016/j.cose.2008.08.003 2009

[5] [5]

Outlier Detection with Autoencoder Ensembles

Chen, Jing, Sathe, Saket, Aggarwal, Charu, and Turaga, Deepak. Outlier Detection with Autoencoder Ensembles. Proceedings of the 2017 SIAM International Conference on Data Mining, pages 90–98, 2017. https://doi. org/10.1137/1.9781611974973.11

work page doi:10.1137/1.9781611974973.11 2017

[6] [6]

Masked Autoregressive Flow for Density Estimation

Papamakarios, George, Pavlakou, Theo, and Murray, Iain. Masked Autoregressive Flow for Density Estimation. Advances in Neural Information Processing Systems (NeurIPS), 30:2335–2344, 2017. https://proceedings. neurips.cc/paper/2017/hash/6c150ed4-Paper.pdf

2017

[7] [7]

A Discriminative Feature Learning Approach for Deep Face Recognition.European Conference on Computer Vision (ECCV), pages 499–515

Wen, Yandong, Zhang, Kaipeng, Li, Zhifeng, and Qiao, Yu. A Discriminative Feature Learning Approach for Deep Face Recognition.European Conference on Computer Vision (ECCV), pages 499–515. Springer, 2016. https://doi.org/10.1007/978-3-319-46478-7_31

doi:10.1007/978-3-319-46478-7_31 2016

[8] [8]

Dimensionality Reduction by Learning an Invariant Mapping

Hadsell, Raia, Chopra, Sumit, and LeCun, Yann. Dimensionality Reduction by Learning an Invariant Mapping. 2006 IEEE Computer Society Conference on Computer Vision and Pattern Recognition (CVPR’06), 2:1735–1742, 2006.https://doi.org/10.1109/CVPR.2006.100

doi:10.1109/cvpr.2006.100 2006

[9] [9]

Schroff, Florian, Kalenichenko, Dmitry, and Philbin, James. FaceNet: A Unified Embedding for Face Recognition and Clustering.Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition (CVPR), pages 815–823, 2015.https://doi.org/10.1109/CVPR.2015.7298682

arXiv 2015

[10] [10]

Wang, Feng, Deng, Jian, Liu, Weiyang, and Wang, C. AMC-Loss: Angular Margin Contrastive Loss for Insightful Representation Learning.Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition Workshops (CVPRW), pages 3659–3666, 2020

2020

[11] [11]

LossTransform: Reformulating Contrastive Objectives for Robust Representation Learning.International Conference on Learning Representations (ICLR), 2025

Liu, Z., Chen, X., and Zhang, H. LossTransform: Reformulating Contrastive Objectives for Robust Representation Learning.International Conference on Learning Representations (ICLR), 2025

2025

[12] [12]

Attention Is All You Need.Advances in Neural Information Processing Systems (NeurIPS), 30:5998–6008, 2017.https://arxiv.org/abs/1706.03762

Vaswani, Ashish, Shazeer, Noam, Parmar, Niki, Uszkoreit, Jakob, Jones, Llion, Gomez, Aidan N, Kaiser, Łukasz, and Polosukhin, Illia. Attention Is All You Need.Advances in Neural Information Processing Systems (NeurIPS), 30:5998–6008, 2017.https://arxiv.org/abs/1706.03762

Pith/arXiv arXiv 2017

[13] [13]

Representation Learning: A Review and New Perspectives

Bengio, Yoshua, Courville, Aaron, and Vincent, Pascal. Representation Learning: A Review and New Perspectives. IEEE Transactions on Pattern Analysis and Machine Intelligence, 35(8):1798–1828, 2013. https://doi.org/ 10.1109/TPAMI.2013.50 20 APREPRINT- DECEMBER30, 2025

doi:10.1109/tpami.2013.50 2013

[14] [14]

A Baseline for Detecting Misclassified and Out-of-Distribution Examples in Neural Networks.Proceedings of the International Conference on Learning Representations (ICLR), 2017

Hendrycks, Dan and Gimpel, Kevin. A Baseline for Detecting Misclassified and Out-of-Distribution Examples in Neural Networks.Proceedings of the International Conference on Learning Representations (ICLR), 2017. https://arxiv.org/abs/1610.02136

Pith/arXiv arXiv 2017

[15] [15]

Sharafaldin, Iman, Lashkari, Arash Habibi, and Ghorbani, Ali A. Toward Generating a New Intrusion Detection Dataset and Intrusion Traffic Characterization.Proceedings of the 4th International Conference on Information Sys- tems Security and Privacy (ICISSP), pages 108–116, 2018. https://doi.org/10.5220/0006639801080116

doi:10.5220/0006639801080116 2018

[16] [16]

A Simple Framework for Contrastive Learning of Visual Representations.International Conference on Machine Learning (ICML), pages 1597–1607

Chen, Ting, Kornblith, Simon, Norouzi, Mohammad, and Hinton, Geoffrey. A Simple Framework for Contrastive Learning of Visual Representations.International Conference on Machine Learning (ICML), pages 1597–1607. PMLR, 2020.https://arxiv.org/abs/2002.05709

Pith/arXiv arXiv 2020

[17] [17]

Decoupled Weight Decay Regularization.International Conference on Learning Representations (ICLR), 2019.https://arxiv.org/abs/1711.05101

Loshchilov, Ilya, and Hutter, Frank. Decoupled Weight Decay Regularization.International Conference on Learning Representations (ICLR), 2019.https://arxiv.org/abs/1711.05101

Pith/arXiv arXiv 2019

[18] [18]

Data Preprocessing for Supervised Learning.International Journal of Computer Science, 1(2):111–117, 2006

Kotsiantis, Sotiris B., Kanellopoulos, Dimitris, and Pintelas, Panayiotis E. Data Preprocessing for Supervised Learning.International Journal of Computer Science, 1(2):111–117, 2006. 21

2006